IVI Notebook

#Setup

#install.packages("tidyverse")
#install.packages("PerformanceAnalytics")
#install.packages("ggfortify")
#install.packages("fastDummies")

library(tidyverse) # core package includes following packages: tidyr, dplyr, ggplot2, readr, purrr, tibble, stringr, forcats
library(plotly)
library("PerformanceAnalytics") #for correlation
library(broom) # for model quantification
library(ggfortify) # for visualizing model fits
library(fastDummies)
library(dplyr)
library(ggplot2)

library(tidyverse)
library(dplyr)
library(tidyr)
library(ggplot2)
library(magrittr)
#install.packages("PerformanceAnalytics")
library("PerformanceAnalytics")
#install.packages("scales")
#library(scales)

Data Wrangling

bike_data <- read_csv("SeoulBikeData.csv",
  show_col_types = FALSE,
  col_types = cols(Date = col_date(format = "%d/%m/%Y"),
    Seasons = col_factor(levels = c("Winter", "Spring", "Summer", "Autumn"),
                         ordered = TRUE),
    Holiday = col_factor(),
    "Functioning Day" = col_factor()
  ))

bike_data <- bike_data %>%
  mutate(day = weekdays(Date), 
         month = months(Date),
         day_time = case_when(
           Hour >= 5 & Hour < 11 ~ "Morning",
           Hour >= 11 & Hour < 15 ~ "Noon",
           Hour >= 15 & Hour < 18 ~ "Afternoon",
           Hour >= 18 & Hour < 22 ~ "Evening",
           Hour < 5 | Hour >= 22 ~ "Night")) %>%
  select(Date,month,day,Hour,day_time, Holiday, 'Rented Bike Count',everything())

bike_data$day_time <- factor(bike_data$day_time,
                             levels = c("Morning", "Noon", "Afternoon", "Evening", "Night"),ordered = TRUE)

bike_data$day <- factor(bike_data$day, 
                        levels = c("Montag", "Dienstag", "Mittwoch", "Donnerstag", "Freitag", "Samstag", "Sonntag" ))

bike_data$month <- factor(bike_data$month, 
                          levels = c("Januar", "Februar", "März", "April", "Mai", "Juni", "Juli", "August", "September", "Oktober","November", "Dezember"))

#LO1: Performance ## Grafik 1

grafik_1 <- bike_data %>%
  plot_ly(x = ~Seasons) %>%
    add_histogram(color = I("darkgreen"), opacity = 0.9) %>%
  layout(title = "Total bike count by seasons")

Zeit messen bei Grafik 1

start <- Sys.time()
grafik_1

end <- Sys.time()
print(end - start)

Time difference of 0.409903 secs

Grafik 2

grafik_2 <- ggplot(bike_data, aes(`Temperature` ,`Rented Bike Count`, color = `Seasons`))+
  geom_jitter(alpha = 0.3)+
  scale_fill_grey(start = 0.2, end = 0.8,na.value = "red")+
  stat_smooth(method = lm, se = FALSE, color = "red")+
  labs(
    x = "Temperature in Celsius",
    y = "Rented Bikes",
    title = "Correlation between temperature and rented bikes")+
  theme_minimal()

Zeit messen bei Grafik 2

start <- Sys.time()
grafik_2

end <- Sys.time()
print(end - start)

Time difference of 1.286342 secs

#LO2: Dashboard design principes ## Grafik 3:

grafik_3 <- bike_data %>%
  plot_ly(x = ~Seasons) %>%
    add_histogram(color = I("navy"), opacity = 0.9) %>%
  layout(title = "Total bike count by seasons")

grafik_3

##Grafik 4:

grafik_4 <- ggplot(bike_data, aes(`Temperature` ,`Rented Bike Count`, color = `Seasons`, colors = "Dark2"))+
  geom_jitter(alpha = 0.3)+
  scale_fill_grey(start = 0.2, end = 0.8,na.value = "red")+
  stat_smooth(method = lm, se = FALSE, color = "black")+
  labs(
    x = "Temperature in Celsius",
    y = "Rented Bikes",
    title = "Correlation between temperature and rented bikes")+
  theme_minimal()

ggplotly(grafik_4)

`geom_smooth()` using formula 'y ~ x'

##Grafik 5

grafik_5 <- ggplot(bike_data,aes(x= Snowfall, y = `Rented Bike Count`))+
  geom_jitter(shape=8, (aes(color = Temperature)))+
  scale_color_gradient(low="dark blue", high= "light blue")+
  facet_wrap(~month)+
  xlab("Schnee in Zentimetern") +
  ylab("Ausgeliehene Fahrräder")+
  ggtitle("Anzahl ausgeliehene Fahrräder und der Einfluss von Schnee")+
  theme_minimal()

grafik_5

ggplotly(grafik_5)

Zoomen und Slider: Schlechtes Beispiel

grafik_6 <- ggplot(bike_data, aes(Hour, `Rented Bike Count`))+
  geom_point(aes(color = day), alpha = 0.5) +
  geom_smooth(aes(color = day, fill = day), method = "lm")+
  xlab("Uhrzeit") +
  ylab("Anzahl ausgeliehener Fahrräder")+
  theme_minimal()

grafik_6

ggplotly(grafik_6, dynamicTicks = TRUE) %>%
  rangeslider() %>%
  layout(hovermode = "x")

`geom_smooth()` using formula 'y ~ x'

Zoomen und Slider: Gutes Beispiel

# Schritt 1: Übersichtliche Grafik machen mit Stunde und ausgeliehenen Fahrrädern nach Wochentag aufgeteilt. 

library(RColorBrewer)
display.brewer.all(colorblindFriendly = TRUE)

grafik_8 <- bike_data %>% 
  ggplot(aes(Hour, `Rented Bike Count`, color = day)) +
  geom_smooth(se = F, size = 2) +
  xlab("Uhrzeit") +
  ylab("Anzahl ausgeliehener Fahrräder")+
  scale_color_discrete("")+
  scale_color_brewer(palette = "Paired")+
  theme_minimal()

Scale for 'colour' is already present. Adding another scale for 'colour', which will replace the existing
scale.

grafik_8

#Schritt 2: Den Slider hinzufügen
ggplotly(grafik_8, dynamicTicks = TRUE) %>%
  rangeslider() %>%
  layout(hovermode = "x", title = "Der Trend von Fahrradausleihen zu verschiedenen Zeiten und an Wochentagen")

`geom_smooth()` using method = 'gam' and formula 'y ~ s(x, bs = "cs")'

#LO3: HCl basics

# Einfacher Linienplot zum Regen in Seoul

ggplot(bike_data) +
  geom_line(aes(x=Date,y=Snowfall),color="deepskyblue")+
  geom_line(aes(x=Date, y=Rainfall), color="navy")+
  labs(title="Täglich gemessener Schneefall und Regen über ein Jahr")+
  xlab(label = "Datum")+ ylab(label = " ")+
  scale_x_date(date_breaks="months",date_labels="%Y-%m")+
  theme_minimal()

rslocator <- function(n=512, type="p", ...)
 {
 on.exit(return(list(x=x,y=y))) # output even when function is canceled with ESC in console
 x <- y <- NULL
 i <- 1
 while(i<=n)
   {
   d <- locator(1)
   if(is.null(d)) break # If user pressed ESC in Rstudio Graphics window
   x <- c(x, d$x)
   y <- c(y, d$y)
   points(x,y, type=type, ...)
   i <- i+1
   }
 }

#LO4: Evaluation